Good advice in general, but a lot of that should be unnecessary in a properly written OS.
Regarding background processes, at least in Mac OS X, most daemons are heavily tuned to not do anything until asked. Instead of polling from user space up into the kernel, they call up into the kernel and then sit waiting in select() on a socket or mach_msg_trap on a mach port until the other end of the socket/pipe/mach port sends a message to tell them that something needs to happen. The same concept extends heavily throughout the OS. Some parts of the OS are even launch-on-demand. For example, if you run a daemon like named, it doesn't start running until the first client request. In addition to reducing boot time, this means that daemons that don't really need to be running never get launched in the first place, reducing memory contention that can cause paging, which is a BIG cause of battery life problems.
The power management goal of any OS should be to keep CPUs in the deepest C state (idle) for as long as possible. This means, among other things, requiring that the CPU reach a certain minimum load level before you schedule anything on an idle CPU core, tuning all your background apps to do as little work as possible, never polling the kernel, increasing the timeouts on select() calls and similar to minimize CPU overhead while daemons and apps are idle, disabling the sync/update daemon entirely after a couple of near-idle periods so that system log messages don't prevent the hard drive from spinning down (or, ideally, switching to a more modern logging daemon that works with the kernel to aggregate log data in RAM when the hard drive needs to go idle), modifying buffer cache code so that transient files (files that live less than a couple of seconds before being deleted) are never written out to disk, and probably a thousand other things that I'm not thinking of right now.
Another thing you mentioned that really shouldn't be necessary is disabling devices. Yes, it can help, but it shouldn't help much for idle devices unless something is very wrong.
The wireless stack should only scan when an application polls it and asks for an updated list of access points, which should only happen when the user clicks on an Airport menu extra or whatever the Linux equivalent is. If the wireless driver is continuously scanning, then the driver or the driver model is fundamentally broken by design and should be fixed. Similarly, the wireless transmit and receive gain *should* be set to the minimum level at which communication is possible until data is received or transmitted, at which point it should increase power to a higher level to maximize throughput until such time as the data stops, at which point it should cut the power back again. And if the card is not associated with a network (and is not in IBSS mode), the card's power state should basically be off.
Likewise, a webcam, assuming that the firmware is written correctly, should be electrically powered off in software until needed. The only thing that should be drawing power when idle is the USB silicon inside the webcam. When you open the device for reading a video stream, the device should be powered up and initialized on demand, and it should be shut down when you close the last read stream. If that isn't happening, it's a flaw in the design of the driver stack or the hardware/firmware, and a fairly glaring one at that.
So yes, you can nickel and dime your way into a tiny bit of extra battery life by following those steps, but if they make a big difference, it means that there's something massively wrong with the OS itself. The improvement should be on the order of a few extra minutes of battery life if power is being managed correctly at the OS level.:-)
If the GPU is doing the decompression, you might even be able to keep the CPU in a C state most of the time (idle/near-idle). Whether this would be a benefit or not depends on how much power the GPU takes to decompress the video.
My point was that depending on how many of those "independent contractors" were really doing the work for their employer versus doing it themselves, it could be up to four times that number.
Yeah, I couldn't believe that wasn't thrown out of the list. What makes it funnier, though, is that the #5 contributor could well be the very set of isolated kernel hackers that they claim don't exist.
Pedantically, I mean nearest to a place where that company has nexus in the state. Unless the company has their legal offices in that part of Texas, the fact that they sometimes sell products to companies who sell them to other companies who sell them to retailers who sell them to people who live in Texas is largely immaterial. Do they transact business in Texas? Only if they have sufficient nexus to be required to collect Texas sales tax. If they aren't charging sales tax there, they shouldn't be suing there.
No surprise, really. For anything other than somebody bombing the cell towers, cellular service is likely to be better in an emergency. The cell system has built-in battery and generator backup for keep most of the main towers up and running, but unlike land lines, there are far fewer such pieces of equipment to keep running, and the number of potential points of failure (e.g. trunk line breaks) are far fewer. Also, in the event of a disaster, IIRC, the towers can be switched into an emergency mode in which calls are only allowed through if they're going to 911 or are to/from known emergency responders. This could potentially allow it to handle emergency calls better than the land line circuits in some cases.
Also, with cellular service, they can quickly bring in emergency cell tower trucks to provide service. Try that with land line infrastructure. These are also handy for augmenting cellular service when the towers have not been damaged, e.g. for a huge political convention, a major sporting event, etc.
That said, if you want reliable communication, you really can't beat ham radio.... None of this addresses the issue of broadband, of course, for which cellular service continues to suck and will likely suck for the foreseeable future thanks to the laws of physics.
I didn't say that companies should always get rid of people. I said that they needed to be agile. Occasionally, you have to lay people off, but there are usually ways of being agile that don't involve significant layoffs. I watched a few years ago as an entire software team of a few dozen employees became unnecessary at my current employer. (The entire technology transitioned rapidly from an active product into maintenance mode after being replaced by a new product.) Their manager found jobs within the company for every single one of his employees before he was laid off. That's good agile management, and the fact that the employees were able to shift to other teams and integrate so easily into those teams is a sign of good hiring of agile employees as well. It also meant that the company was stronger for having kept those employees because it retained the expertise in that maintenance-mode product, should it require more complex maintenance than expected. (The product in question is now completely dead, but the majority of those folks are still with the company, some having changed jobs again since then.)
I've been lucky enough to work for two companies that have been fairly agile. If you have never seen it in action, I can understand why you'd be reticent. Agile employment means being able to shift people to other tasks quickly. It means being able to borrow a third of your product development team to work on a new product and adjusting schedules as needed to make it happen. It means that next week, they could tell me instead of writing API documentation, I might be writing user documentation. Moreover, it means that next week, instead of writing about software development, I could actually be writing code that makes it into the product. As a tech writer, I've contributed code patches to fairly critical components. Agile employment means hiring people who are capable of shifting gears (often radically), who are intelligent enough to rapidly adapt to changing needs, and who are willing to adjust to changing economic times in order to be as productive as possible (within reason).
It's not at all about getting rid of lots of people, though it does require being able to get rid of people who turn out to be incompetent---another thing that union contracts tend to make particularly difficult. Mainly, though, it means hiring people who are wiling to learn new skills and having contracts that do not deliberately stand in the way of employees learning new skills. From a union's perspective, the theory is that by preventing people from picking up new skill sets, it decreases the risk of consolidating two positions into one. The reality, however, is that it makes the company progressively less agile, which eventually leads to inevitable financial collapse.
The absence of DNA is generally not useful in any way, but finding someone else's DNA under the fingernails of the murdered girl is quite often useful in proving someone's innocence, or at least establishing substantial doubt. The child of the rape victim turning out not to be yours can be similarly useful. And so on. DNA can quite often prove innocence by strongly suggesting that someone else was involved, thus making a case of mistaken identity seem like the most plausible explanation.
It's not that the unions want more money for their employees. If it were mainly about money and benefits, they wouldn't be nearly as problematic for big industry. The problem with unions in industry is their protectionist policies that A. don't allow the manufacturers to easily cut jobs that are no longer needed and hire in other areas where more people are needed, B. require extended apprenticeship periods under other union members instead of allowing the company itself to retrain employees to handle new jobs, and C. contractually specify overspecialization of employees for a very specific task such that if an employee gets sick and another one with that highly specific training isn't available, the entire line shuts down and all of the other forty employees sit around drinking beer or whatever.
Such protectionist policies are directly contrary to the basic properties of a successful employer. They don't allow the sort of rapid retooling and retraining that is necessary for a business to compete in a changing economy with changing needs. They result in a company that is bloated and legacy-bound, a company that cannot adapt quickly to changes in requirements, a company that is not nimble, if you will. The result is that companies without such limitations are able to easily innovate, adapt, and change directions on a dime, leaving these slow, lumbering, unionized dinosaurs to bleed red as they play catch-up year after year. In the long run, all those policies do is ensure that the employees are protected right out of a job. An agile employer is a long-term employer even if it is a rat, while a slow, lumbering, triceratops-like employer is doomed to extinction.
True, but at $0.37/kWh, a mere 5 extra watts translates to $16 extra per year per device (most of it probably from coal with my luck). That adds up pretty quickly.
Just to clarify, in both examples, I was referring to the device, not the power cord. Thin DC power cords break. That's why I'm of the opinion that the government should mandate that all power cords on all devices have a connector on both ends and be available as a replacement part without buying an entire power supply.:-) IMHO, the key to good power connector design is to ensure that if something is going to break, it's the cheap power cord and not the expensive device plugged into that power cord.
Regarding the soldering on the Sony Ericsson connectors, the connectors I've seen are a series of metal plates that appear to be a fairly integral part of one section of the phone's plastic case. Assuming they didn't do something stupid in the actual design of the connector, it should be darn near impossible to break without breaking the case. I wouldn't put it past Sony to do something stupid inside the connector, of course, but my SE phone is still functioning (albeit replaced by an iPhone) after about seven or eight years. That's not half bad for a cheap phone.
My point was that somebody will get the bright idea to start advertising wired chargers a few years later as the "green, money-saving alternative to wireless chargers", at which point customers would have at least some of the information....
Wireless power for most consumer devices is without a doubt the most idiotic concept I've heard suggested in the consumer space lately. The EM spectrum is already an awful mess. We shouldn't be raising the noise floor and wrecking our environment by increasing the consumption of all these current-guzzling power supplies just so that people don't have to plug things in. We should be mandating that all traditional transformer-based supplies be replaced with power-factor-corrected switched mode power supplies, not moving one side of the mains transformer farther away from the other side and cutting the efficiency further.... If you really want to simplify things, don't replace the cable with a wireless hack; do what the EU is doing---move towards a standard power supply voltage and connector for small consumer electronics.
We kind of do have that ability in the wireless world. The big difference is the infrastructure cost. You can drop one cell tower for $75,000-$200,000, depending on location, and it can potentially service a 25-30 mile radius. If you stick it on the roof of a building, you can have it up in days. In rural areas, fiber is going to cost you on the order of $16,000 per mile, so if you have one customer at the edge of that 30 mile radius, it's going to cost half a million dollars to reach that one customer. It's easy to see why having multiple wired providers has a lot more trouble with infrastructure costs than wireless, and even with wireless (where competition is theoretically allowed everywhere), there are still many places that are not covered by all of the major wireless providers, and some places that are not covered by any of them....
This just further illustrates why the government should build out wire line infrastructure. If companies want to augment that with their own, fine, but municipal infrastructure should be ubiquitous instead of just being in a handful of cities and towns as it is now.
I would probably tend to make DNS lookups occur at the highest priority, followed by short requests (most HTTP requests, short mail messages, etc.), with longer HTTP or FTP uploads/downloads prioritized similarly to BitTorrent. Basically, the longer it takes, the less the user will generally care how long it takes.
WIth such deeper shaping, however, must come a caveat: transfer speeds must be marketed based on typical continuous download speeds, not based on peak speeds. They don't rank cars by peak MPG while drafting a truck on a long interstate with a tailwind. Internet bandwidth should be rated similarly.... The Comcasts of the world would hate this.:-)
Only poorly designed connectors have this problem. Take a look at Apple's MagSafe connector or the connector that Sony Ericsson uses for examples of how to build connectors that should never wear out in this way, abuse notwithstanding.
And in ten years, when someone points out that you can save $100 per year on electricity by using an all new cabled cell phone charger, suddenly the average user will love cables all over again. The average user doesn't know what he/she wants or needs and never will. Therefore, you should design products to cater to power users in terms of capabilities and complete idiots in terms of ease of use, and ignore the protestations of the proles. Anything else will inevitably lead to products designed by committee that satisfy no one at all.
I've heard that theory put forth by libertarians many times, and it is as wrong now as it has been every other time. There are two very fundamental problems with that theory:
The damage that could be done by allowing anybody to spend a few bucks and dig up roads, driveways, and right-of-way areas that the city or homeowner has to pay to repair is nontrivial. The reason for these limits is that installing new cables is a very invasive process for residents. It's not about limiting competition. It's about limiting disruption.
Even if you completely opened it up to competition, very few communities would ever successfully have competition.
That last one bears explaining. A few years ago, I watched a new cable company try to set up shop in a small university town of about 10,000 people. Here's what happened.
The original cable company is an entrenched business. Regardless of monopoly status, it has been around for years and owns all its own lines. It has no debts because the lines are paid off already. Therefore, its only costs are buying the service from upstream, line maintenance (minimal), handing payments (most of which is done by mail sent to/from a regional office somewhere), and sending people out to connect/disconnect customers and swap out cable boxes. In short, it is largely a cash cow, and has huge profit margins built in.
The new company has to put in tens or hundreds of thousands of dollars worth of cable, equipment, etc. It now has a huge debt. It also has to compete with the existing cable company. It must either do so by providing more channels or undercutting them on price. Unfortunately, because of the construction debt, it must make a certain amount of profit just to stay in business.
The result is that the new company undercuts the entrenched company and makes them angry. The entrenched company undercuts them far enough that they cannot compete and still pay off their construction debt. In spite of taking over a third of the entrenched company's business, after five years, the new company is still hemorrhaging money. Thus, it gives up, declares bankruptcy if needed, and sells all of the new equipment to the entrenched cable company. The entrenched cable company then raises rates to make up the money it lost while competing with the now defunct new company, all the while enjoying the lower maintenance costs of the new equipment that it bought for pennies on the dollar.
And this, my friends, is what inevitably occurs when a business with such huge startup costs tries to compete in a fixed-size market. There is truly no way to prevent this except to take the startup costs out of the picture, either by the government giving a colossal grant to the cable company to cover its infrastructure costs or by the government building the infrastructure to begin with and leasing it out to multiple competitors.
The only way telecom competition can work is if the infrastructure provider and the data provider are not the same company---if the infrastructure provider leases access to the data provider on a nondiscriminatory basis. The government is an ideal builder of infrastructure because it can afford to build it and run it at cost instead of making a profit. Therefore, the ideal form of telecom competition is one in which the government rolls out the fiber and leases fiber access to half a dozen telcos. Everywhere that has done this has seen incredible competition in the telecom space. Most communities that have not done this have little to no competition even if they are completely willing to allow multiple telcos or cable companies to do business in the area. At best, they have partial competition in which the government forces the incumbent telcos to lease access to the lines (e.g. DSL competition).
How do you figure that traffic shaping is good when the ISP has no idea what the traffic is used for? Case in point: I work for an IT shop that supports many physicians offices. one of the primary methods of moving data between offices and hospitals is through EMR applications that USE FTP. Who is the ISP to tell me that my FTP traffic is less important than Disney's HTTP traffic?
HTTP traffic, no. VoIP traffic, maybe---not more important, but more time sensitive. QoS is reasonable for limiting latency for time-critical payloads. Apart from that, QoS should never occur. If an ISP is doing QoS to prioritize bulk traffic relative to other bulk traffic, it is doing something wrong. All bulk traffic should be inherently equal.
Neither. I would follow the traditional rules for choice of venue, and thus would sue in a court near me. Normally, the venue for a lawsuit should be the court nearest to the plaintiff or the court nearest to the defendant.
On the other hand, pretty much the entire computer industry has held the Marshall, TX district courts in utter contempt for at least a decade. It's the patent troll venue of choice because it is so favorable to patent trolls. They've decided incorrectly on nearly every patent troll case that has ever been posted on Slashdot. Note that they are almost inevitably filed in that district. When you see a few wrong decisions from a district, it's a fluke. When you see consistently wrong decisions made by a district and see almost every wrong decision of a certain genre coming out of a single district, it's time to start working on getting those corrupt judges impeached from the bench. It's a kangaroo court, and it has no business existing in these United States.
If I were one of Microsoft's lawyers, I'd do everything I could to try to push the judge's buttons as much as possible, both because it's fun to mess with idiots and because if they can push the buttons enough to get the judge to screw up massively in a way that shows a gross bias against all patent defendants (as this particular court appears to have), IIRC, that can make the appeals process easier (not the trial, just getting a trial in the first place) and might even make it possible to get the verdict set aside and force a change of venue to a more reasonable court.
They may not be patent trolls, but they sure are acting like patent trolls. They applied for a patent on something that has been done for decades, then chose to sue in a district in Texas that is known for automatically allowing all patents and ignoring validity challenges (and thus is the venue of choice for patent trolls). About the only thing not patent troll about them is that they actually had a product at one time. Even still, if it looks like a duck and quacks like a duck....
I'd go with the "effective complete copies" theory, i.e. if you give 1/100th of the file to 1000 people, you gave out the effective equivalent of 10 compete copies. Unfortunately, I think the DOJ seems to believe this is equivalent to distributing 1,000 complete copies.... This is what happens when you put a bunch of RIAA lawyers in positions of power....
Slashdot Mirror
But it would really annoy 3ttsburgh.
I don't know who this Mr. Ethics is, but Mr. Morales is moving up in the world....
Good advice in general, but a lot of that should be unnecessary in a properly written OS.
Regarding background processes, at least in Mac OS X, most daemons are heavily tuned to not do anything until asked. Instead of polling from user space up into the kernel, they call up into the kernel and then sit waiting in select() on a socket or mach_msg_trap on a mach port until the other end of the socket/pipe/mach port sends a message to tell them that something needs to happen. The same concept extends heavily throughout the OS. Some parts of the OS are even launch-on-demand. For example, if you run a daemon like named, it doesn't start running until the first client request. In addition to reducing boot time, this means that daemons that don't really need to be running never get launched in the first place, reducing memory contention that can cause paging, which is a BIG cause of battery life problems.
The power management goal of any OS should be to keep CPUs in the deepest C state (idle) for as long as possible. This means, among other things, requiring that the CPU reach a certain minimum load level before you schedule anything on an idle CPU core, tuning all your background apps to do as little work as possible, never polling the kernel, increasing the timeouts on select() calls and similar to minimize CPU overhead while daemons and apps are idle, disabling the sync/update daemon entirely after a couple of near-idle periods so that system log messages don't prevent the hard drive from spinning down (or, ideally, switching to a more modern logging daemon that works with the kernel to aggregate log data in RAM when the hard drive needs to go idle), modifying buffer cache code so that transient files (files that live less than a couple of seconds before being deleted) are never written out to disk, and probably a thousand other things that I'm not thinking of right now.
Another thing you mentioned that really shouldn't be necessary is disabling devices. Yes, it can help, but it shouldn't help much for idle devices unless something is very wrong.
The wireless stack should only scan when an application polls it and asks for an updated list of access points, which should only happen when the user clicks on an Airport menu extra or whatever the Linux equivalent is. If the wireless driver is continuously scanning, then the driver or the driver model is fundamentally broken by design and should be fixed. Similarly, the wireless transmit and receive gain *should* be set to the minimum level at which communication is possible until data is received or transmitted, at which point it should increase power to a higher level to maximize throughput until such time as the data stops, at which point it should cut the power back again. And if the card is not associated with a network (and is not in IBSS mode), the card's power state should basically be off.
Likewise, a webcam, assuming that the firmware is written correctly, should be electrically powered off in software until needed. The only thing that should be drawing power when idle is the USB silicon inside the webcam. When you open the device for reading a video stream, the device should be powered up and initialized on demand, and it should be shut down when you close the last read stream. If that isn't happening, it's a flaw in the design of the driver stack or the hardware/firmware, and a fairly glaring one at that.
So yes, you can nickel and dime your way into a tiny bit of extra battery life by following those steps, but if they make a big difference, it means that there's something massively wrong with the OS itself. The improvement should be on the order of a few extra minutes of battery life if power is being managed correctly at the OS level. :-)
If the GPU is doing the decompression, you might even be able to keep the CPU in a C state most of the time (idle/near-idle). Whether this would be a benefit or not depends on how much power the GPU takes to decompress the video.
My point was that depending on how many of those "independent contractors" were really doing the work for their employer versus doing it themselves, it could be up to four times that number.
Yeah, I couldn't believe that wasn't thrown out of the list. What makes it funnier, though, is that the #5 contributor could well be the very set of isolated kernel hackers that they claim don't exist.
Pedantically, I mean nearest to a place where that company has nexus in the state. Unless the company has their legal offices in that part of Texas, the fact that they sometimes sell products to companies who sell them to other companies who sell them to retailers who sell them to people who live in Texas is largely immaterial. Do they transact business in Texas? Only if they have sufficient nexus to be required to collect Texas sales tax. If they aren't charging sales tax there, they shouldn't be suing there.
No surprise, really. For anything other than somebody bombing the cell towers, cellular service is likely to be better in an emergency. The cell system has built-in battery and generator backup for keep most of the main towers up and running, but unlike land lines, there are far fewer such pieces of equipment to keep running, and the number of potential points of failure (e.g. trunk line breaks) are far fewer. Also, in the event of a disaster, IIRC, the towers can be switched into an emergency mode in which calls are only allowed through if they're going to 911 or are to/from known emergency responders. This could potentially allow it to handle emergency calls better than the land line circuits in some cases.
Also, with cellular service, they can quickly bring in emergency cell tower trucks to provide service. Try that with land line infrastructure. These are also handy for augmenting cellular service when the towers have not been damaged, e.g. for a huge political convention, a major sporting event, etc.
That said, if you want reliable communication, you really can't beat ham radio.... None of this addresses the issue of broadband, of course, for which cellular service continues to suck and will likely suck for the foreseeable future thanks to the laws of physics.
I didn't say that companies should always get rid of people. I said that they needed to be agile. Occasionally, you have to lay people off, but there are usually ways of being agile that don't involve significant layoffs. I watched a few years ago as an entire software team of a few dozen employees became unnecessary at my current employer. (The entire technology transitioned rapidly from an active product into maintenance mode after being replaced by a new product.) Their manager found jobs within the company for every single one of his employees before he was laid off. That's good agile management, and the fact that the employees were able to shift to other teams and integrate so easily into those teams is a sign of good hiring of agile employees as well. It also meant that the company was stronger for having kept those employees because it retained the expertise in that maintenance-mode product, should it require more complex maintenance than expected. (The product in question is now completely dead, but the majority of those folks are still with the company, some having changed jobs again since then.)
I've been lucky enough to work for two companies that have been fairly agile. If you have never seen it in action, I can understand why you'd be reticent. Agile employment means being able to shift people to other tasks quickly. It means being able to borrow a third of your product development team to work on a new product and adjusting schedules as needed to make it happen. It means that next week, they could tell me instead of writing API documentation, I might be writing user documentation. Moreover, it means that next week, instead of writing about software development, I could actually be writing code that makes it into the product. As a tech writer, I've contributed code patches to fairly critical components. Agile employment means hiring people who are capable of shifting gears (often radically), who are intelligent enough to rapidly adapt to changing needs, and who are willing to adjust to changing economic times in order to be as productive as possible (within reason).
It's not at all about getting rid of lots of people, though it does require being able to get rid of people who turn out to be incompetent---another thing that union contracts tend to make particularly difficult. Mainly, though, it means hiring people who are wiling to learn new skills and having contracts that do not deliberately stand in the way of employees learning new skills. From a union's perspective, the theory is that by preventing people from picking up new skill sets, it decreases the risk of consolidating two positions into one. The reality, however, is that it makes the company progressively less agile, which eventually leads to inevitable financial collapse.
The absence of DNA is generally not useful in any way, but finding someone else's DNA under the fingernails of the murdered girl is quite often useful in proving someone's innocence, or at least establishing substantial doubt. The child of the rape victim turning out not to be yours can be similarly useful. And so on. DNA can quite often prove innocence by strongly suggesting that someone else was involved, thus making a case of mistaken identity seem like the most plausible explanation.
It's not that the unions want more money for their employees. If it were mainly about money and benefits, they wouldn't be nearly as problematic for big industry. The problem with unions in industry is their protectionist policies that A. don't allow the manufacturers to easily cut jobs that are no longer needed and hire in other areas where more people are needed, B. require extended apprenticeship periods under other union members instead of allowing the company itself to retrain employees to handle new jobs, and C. contractually specify overspecialization of employees for a very specific task such that if an employee gets sick and another one with that highly specific training isn't available, the entire line shuts down and all of the other forty employees sit around drinking beer or whatever.
Such protectionist policies are directly contrary to the basic properties of a successful employer. They don't allow the sort of rapid retooling and retraining that is necessary for a business to compete in a changing economy with changing needs. They result in a company that is bloated and legacy-bound, a company that cannot adapt quickly to changes in requirements, a company that is not nimble, if you will. The result is that companies without such limitations are able to easily innovate, adapt, and change directions on a dime, leaving these slow, lumbering, unionized dinosaurs to bleed red as they play catch-up year after year. In the long run, all those policies do is ensure that the employees are protected right out of a job. An agile employer is a long-term employer even if it is a rat, while a slow, lumbering, triceratops-like employer is doomed to extinction.
Odd. I don't believe I've ever seen a filling station in the U.S. that didn't sell diesel fuel. Where in the world are you?
True, but at $0.37/kWh, a mere 5 extra watts translates to $16 extra per year per device (most of it probably from coal with my luck). That adds up pretty quickly.
Just to clarify, in both examples, I was referring to the device, not the power cord. Thin DC power cords break. That's why I'm of the opinion that the government should mandate that all power cords on all devices have a connector on both ends and be available as a replacement part without buying an entire power supply. :-) IMHO, the key to good power connector design is to ensure that if something is going to break, it's the cheap power cord and not the expensive device plugged into that power cord.
Regarding the soldering on the Sony Ericsson connectors, the connectors I've seen are a series of metal plates that appear to be a fairly integral part of one section of the phone's plastic case. Assuming they didn't do something stupid in the actual design of the connector, it should be darn near impossible to break without breaking the case. I wouldn't put it past Sony to do something stupid inside the connector, of course, but my SE phone is still functioning (albeit replaced by an iPhone) after about seven or eight years. That's not half bad for a cheap phone.
My point was that somebody will get the bright idea to start advertising wired chargers a few years later as the "green, money-saving alternative to wireless chargers", at which point customers would have at least some of the information....
Wireless power for most consumer devices is without a doubt the most idiotic concept I've heard suggested in the consumer space lately. The EM spectrum is already an awful mess. We shouldn't be raising the noise floor and wrecking our environment by increasing the consumption of all these current-guzzling power supplies just so that people don't have to plug things in. We should be mandating that all traditional transformer-based supplies be replaced with power-factor-corrected switched mode power supplies, not moving one side of the mains transformer farther away from the other side and cutting the efficiency further.... If you really want to simplify things, don't replace the cable with a wireless hack; do what the EU is doing---move towards a standard power supply voltage and connector for small consumer electronics.
We kind of do have that ability in the wireless world. The big difference is the infrastructure cost. You can drop one cell tower for $75,000-$200,000, depending on location, and it can potentially service a 25-30 mile radius. If you stick it on the roof of a building, you can have it up in days. In rural areas, fiber is going to cost you on the order of $16,000 per mile, so if you have one customer at the edge of that 30 mile radius, it's going to cost half a million dollars to reach that one customer. It's easy to see why having multiple wired providers has a lot more trouble with infrastructure costs than wireless, and even with wireless (where competition is theoretically allowed everywhere), there are still many places that are not covered by all of the major wireless providers, and some places that are not covered by any of them....
This just further illustrates why the government should build out wire line infrastructure. If companies want to augment that with their own, fine, but municipal infrastructure should be ubiquitous instead of just being in a handful of cities and towns as it is now.
I would probably tend to make DNS lookups occur at the highest priority, followed by short requests (most HTTP requests, short mail messages, etc.), with longer HTTP or FTP uploads/downloads prioritized similarly to BitTorrent. Basically, the longer it takes, the less the user will generally care how long it takes.
WIth such deeper shaping, however, must come a caveat: transfer speeds must be marketed based on typical continuous download speeds, not based on peak speeds. They don't rank cars by peak MPG while drafting a truck on a long interstate with a tailwind. Internet bandwidth should be rated similarly.... The Comcasts of the world would hate this. :-)
Only poorly designed connectors have this problem. Take a look at Apple's MagSafe connector or the connector that Sony Ericsson uses for examples of how to build connectors that should never wear out in this way, abuse notwithstanding.
And in ten years, when someone points out that you can save $100 per year on electricity by using an all new cabled cell phone charger, suddenly the average user will love cables all over again. The average user doesn't know what he/she wants or needs and never will. Therefore, you should design products to cater to power users in terms of capabilities and complete idiots in terms of ease of use, and ignore the protestations of the proles. Anything else will inevitably lead to products designed by committee that satisfy no one at all.
I've heard that theory put forth by libertarians many times, and it is as wrong now as it has been every other time. There are two very fundamental problems with that theory:
That last one bears explaining. A few years ago, I watched a new cable company try to set up shop in a small university town of about 10,000 people. Here's what happened.
The original cable company is an entrenched business. Regardless of monopoly status, it has been around for years and owns all its own lines. It has no debts because the lines are paid off already. Therefore, its only costs are buying the service from upstream, line maintenance (minimal), handing payments (most of which is done by mail sent to/from a regional office somewhere), and sending people out to connect/disconnect customers and swap out cable boxes. In short, it is largely a cash cow, and has huge profit margins built in.
The new company has to put in tens or hundreds of thousands of dollars worth of cable, equipment, etc. It now has a huge debt. It also has to compete with the existing cable company. It must either do so by providing more channels or undercutting them on price. Unfortunately, because of the construction debt, it must make a certain amount of profit just to stay in business.
The result is that the new company undercuts the entrenched company and makes them angry. The entrenched company undercuts them far enough that they cannot compete and still pay off their construction debt. In spite of taking over a third of the entrenched company's business, after five years, the new company is still hemorrhaging money. Thus, it gives up, declares bankruptcy if needed, and sells all of the new equipment to the entrenched cable company. The entrenched cable company then raises rates to make up the money it lost while competing with the now defunct new company, all the while enjoying the lower maintenance costs of the new equipment that it bought for pennies on the dollar.
And this, my friends, is what inevitably occurs when a business with such huge startup costs tries to compete in a fixed-size market. There is truly no way to prevent this except to take the startup costs out of the picture, either by the government giving a colossal grant to the cable company to cover its infrastructure costs or by the government building the infrastructure to begin with and leasing it out to multiple competitors.
The only way telecom competition can work is if the infrastructure provider and the data provider are not the same company---if the infrastructure provider leases access to the data provider on a nondiscriminatory basis. The government is an ideal builder of infrastructure because it can afford to build it and run it at cost instead of making a profit. Therefore, the ideal form of telecom competition is one in which the government rolls out the fiber and leases fiber access to half a dozen telcos. Everywhere that has done this has seen incredible competition in the telecom space. Most communities that have not done this have little to no competition even if they are completely willing to allow multiple telcos or cable companies to do business in the area. At best, they have partial competition in which the government forces the incumbent telcos to lease access to the lines (e.g. DSL competition).
HTTP traffic, no. VoIP traffic, maybe---not more important, but more time sensitive. QoS is reasonable for limiting latency for time-critical payloads. Apart from that, QoS should never occur. If an ISP is doing QoS to prioritize bulk traffic relative to other bulk traffic, it is doing something wrong. All bulk traffic should be inherently equal.
Neither. I would follow the traditional rules for choice of venue, and thus would sue in a court near me. Normally, the venue for a lawsuit should be the court nearest to the plaintiff or the court nearest to the defendant.
On the other hand, pretty much the entire computer industry has held the Marshall, TX district courts in utter contempt for at least a decade. It's the patent troll venue of choice because it is so favorable to patent trolls. They've decided incorrectly on nearly every patent troll case that has ever been posted on Slashdot. Note that they are almost inevitably filed in that district. When you see a few wrong decisions from a district, it's a fluke. When you see consistently wrong decisions made by a district and see almost every wrong decision of a certain genre coming out of a single district, it's time to start working on getting those corrupt judges impeached from the bench. It's a kangaroo court, and it has no business existing in these United States.
If I were one of Microsoft's lawyers, I'd do everything I could to try to push the judge's buttons as much as possible, both because it's fun to mess with idiots and because if they can push the buttons enough to get the judge to screw up massively in a way that shows a gross bias against all patent defendants (as this particular court appears to have), IIRC, that can make the appeals process easier (not the trial, just getting a trial in the first place) and might even make it possible to get the verdict set aside and force a change of venue to a more reasonable court.
They may not be patent trolls, but they sure are acting like patent trolls. They applied for a patent on something that has been done for decades, then chose to sue in a district in Texas that is known for automatically allowing all patents and ignoring validity challenges (and thus is the venue of choice for patent trolls). About the only thing not patent troll about them is that they actually had a product at one time. Even still, if it looks like a duck and quacks like a duck....
I'd go with the "effective complete copies" theory, i.e. if you give 1/100th of the file to 1000 people, you gave out the effective equivalent of 10 compete copies. Unfortunately, I think the DOJ seems to believe this is equivalent to distributing 1,000 complete copies.... This is what happens when you put a bunch of RIAA lawyers in positions of power....